The Agent Skills Directory

[COMMAND_EXECUTION]: The skill facilitates the execution of a local utility script src/hack.ts via the bun runtime. Executing scripts within the workspace is a standard development task, but the logic within hack.ts is not defined in the skill files.
[INDIRECT_PROMPT_INJECTION]: The skill provides an interface to run workspace-level commands which could ingest untrusted content from the codebase.
Ingestion points: Standard output and error streams from pnpm and bun commands are read by the agent.
Boundary markers: Absent. The agent is not instructed to ignore potential instructions embedded in the command output.
Capability inventory: Subprocess execution for pnpm and bun commands as defined in the Quick Reference section of SKILL.md.
Sanitization: Absent. Data from command execution is passed to the agent without filtering or escaping.

workspace-commands